Natural rocks may be chemically transformed thanks to the operation of pervading fluids in disequilibrium with them. For a given example, a series of zones with distinct mineralogy may develop in space, starting from the initial rock and going to completely new rocks. We will consider the entire set of transformation zones as a whole and call it « system of metasomatic zones ». At the reaction fronts that separate the different zones, chemical components are exchanged between the solid and the fluid; the exchanges occur at the same time and the same place for all components, and the variations of all concentrations are thus correlated. Let take this as granted, and add the condition that a local equilibrium is achieved between the solid and the fluid: we then reach the conclusion that the system of zones as a whole is a connected system. We propose to generalize to this system the concept of variance : let it be the number of intensive parameters that one is able to fix arbitrarily to the inlet fluid and/or to the starting rock without modifying the number and nature of the zones. We show that the variance of a system of metasomatic zones is given by v = c + z -  - 1 where c is the number of independent chemical components, z the number of zones and  the total number of phases, counted as many times as number of zones where present. This rule sets constraints on the difference between the number of phases and the number of zones and on the number of inert and mobile components. On that respect, it brings an improvement with respect to the rule on open systems (Korzhinskii) that could make one think that the mobility property is a local property whereas it is dependent on the whole system. We can say that  - z  ci - 1 where ci is the number of inert components. In addition to its quantitative aspects, our rule merely expresses that the system of metasomatic zones is the result of a reaction between the inlet fluid and the starting material and combine influences of both. It brings an upper limit to the number of parameters that one can decide to fix from outside of the system when studying metasomatic systems. The total of arbitraries is always lower than c - 1, and not 2(c - 1) if the system was not bound, and the arbitraries must be distributed between the inlet fluid and the starting rock. Each specific case needs a specific discussion. The rule also expresses the seeming upstream influence of the starting rock. In the paper, the expression of v is given depending on the inert/mobile status of the components. An example of its use to the case of Costabonne skarns (Pyrenees) is given.